Good Judgments Workshop
Good Judgments Workshop is a 45-minute facilitated session I hosted for my team on not letting the LLMs do the deciding — and an experiment in building the presentation itself with an LLM on Reveal.js instead of traditional tools like PowerPoint or Keynote.
Open the workshop deck · Source on GitHub
Highlights
- Slides as code: text-first semantic sections — diffable in PRs, versioned by git, deep-linkable per slide via hash routing, deployable as a website.
- LLM-authored deck: the source article, audience, and arc went in; 18 structured slides, consistent visual system, and per-slide speaker notes with facilitation timings came out.
- Self-contained Reveal.js 6: the distribution is vendored into the repo — no Node build, no CDN;
python -m http.server 8000is the entire install. - Custom presenter tooling: speaker-notes panel toggled with
N,aria-liveslide status, fixed 1440×810 design surface that scales to any screen. - The content: a five-step judgment loop (Frame → Compare → Prove → Constrain → Own) and a six-question review checklist — Problem, Boundary, Evidence, Smallest move, Signals, Recovery.
- A 12-minute group exercise: review an AI-generated PR smuggling fleet-wide automation, elevated access, and a policy bypass as “faster deployments.”